Feature Dimension Reduction for Microarray Data Analysis Using Locally Linear Embedding

Abstract

Cancer classification is one major application of microarray data analysis. Due to the ultra high dimensionality nature of microarray data, data dimension reduction has drawn special attention for such type of data analysis. The currently available data dimension reduction methods are either supervised, where data need to be labeled, or computational complex. In this paper, we proposed to use a revised locally linear embedding(LLE) method, which is purely unsupervised and fast as the feature extraction strategy for microarray data analysis. Three public available microarray datasets have been used to test the proposed method. The effectiveness of LLE is evaluated by the classification accuracy of a SVM classifier. Generally, the results are promising.

A discernibility-based approach to feature selection for microarray data

“…All of them were taken from [11] and are representative of a class of bioinformatics applications that employ microarray data. Also, these datasets are often used in the literature for feature selection [1][2][3][4][5][6][7][8][9]. Their characteristics are shown in Table I (this table also presents the dimensionality of the reduced feature sets that will be discussed in the next subsection).…”

“…In datasets of high dimensionality, particularly those encountered in bioinformatics applications, feature selection is not merely a plausible but often an essential part of the classification process. To this end, over the last few years a series of methods for feature selection have been introduced, most of them focusing on microarray data [1][2][3][4][5][6][7][8][9], where the number of data points is small whilst the dimensionality of the features is high. In addition, sometimes the accuracy rate is increased when the classifier is applied on the reduced feature set.…”

“…The methods used in the literature have been applied mostly for classification with SVMs [1,2,3], although other approaches have been also considered [4,5,6,7,8,9]. The focus of feature selection on SVM classifiers is not random.…”

“…As regards the methods themselves, feature selection has been accomplished using Genetic Algorithms [1,4], Locally Linear Embedding (LLE) [2], Recursive Salient Analysis (RSA) [3], the selection of a support set [5], a combination of PCA and the UKW clustering algorithm [6], t-tests [4,7], Regression Splines (MARS) [7], Classification and Regression Trees (CART) [7], Random Forests [7], Linear Genetic Programs (LGP) [7], Neural Networks as a similarity measure [8], and clustering analysis [9].…”

A Novel Center Point Initialization Technique for K-means Clustering Algorithm

“…Yeung and Ruzzo (2000) presented an empirical exploration with principal component analysis for grouping gene expression datasets, still the initial center points were also selected here at random. Chao and Chen (2005) as well presented an approach regarding dimensions reduction for microarray data exploration employing Locally Linear Embedding.…”